基于子集模拟和概率失效分析的边坡可靠度分析及风险评估

With the rapid development of computer technology and commonly available personal computers, computer simulation - based methods, such as Monte Carlo Simulation (MCS), is gaining popularity in engineering reliability analysis and risk assessment. MCS, however, suffers from two well-recognized disadvantages: (1) lack of resolution and efficiency at small probability levels; and (2) no insight into the relative contributions of various uncertainties to failure probability. This project aims to mitigate these two disadvantages in reliability analysis of slope stability with an efficient MCS method called Subset Simulation and probabilistic failure analysis. The project contains three major research components: (1) development of a Subset Simulation - based probabilistic slope stability analysis approach for improving the efficiency of MCS; (2) development of a probabilistic failure analysis approach for evaluating, both qualitatively and quantitatively, the effect of various uncertainties on failure probability; (3) using the developed probabilistic slope stability analysis approach and probabilistic failure analysis approach to explore systematically the effects of spatial variability of soil properties and uncertainty on critical slip surface on probability of slope failure. This project will not only lead to significant theoretical development in engineering reliability analysis and slope stability analysis, but also will be beneficial to mitigation of landslide hazards, development of early warning system for landslides, and many major civil infrastructure projects that are currently undergoing in China.

随着计算机应用的普及和蒙特卡洛法在工程可靠度分析及风险评估中的广泛应用，蒙特卡洛法的计算效率低及不确定性传递机理不清的缺点也越来越受关注。本项目针对该问题，结合边坡稳定问题和随机场理论，首先，建立一种基于子集模拟的高效蒙特卡洛法，进行边坡可靠度分析及风险评估，以提高其计算效率；其次，开创性地提出概率失效分析的理论和方法，定性和定量地揭示蒙特卡洛法中不确定性的传递机理；并在此基础上，系统地、定量地探讨土性的空间变异性及临界滑裂面的不确定性对边坡失稳概率的影响。本项目的开展，将推动工程可靠度分析理论的进一步发展，并在完善边坡稳定的分析理论、深化对滑坡机理的认识、滑坡治理以及滑坡预报等方面有着重要的理论价值和科学意义，同时对指导工程设计实践和保障当前各项大型工程建设的胜利完成也有着十分重要的实际工程应用价值。

随着计算机应用的普及和蒙特卡洛法在工程可靠度分析及风险评估中的广泛应用，蒙特卡洛法的计算效率低及不确定性传递机理不清的缺点也越来越受关注。本项目针对该问题，结合边坡稳定问题和随机场理论，首先，建立一种基于子集模拟的高效蒙特卡洛法，进行边坡可靠度分析及风险评估，以提高其计算效率；其次，开创性地提出概率失效分析的理论和方法，定性和定量地揭示蒙特卡洛法中不确定性的传递机理；并在此基础上，系统地、定量地探讨土性的空间变异性及临界滑裂面的不确定性对边坡失稳概率的影响。本项目进一步以微软的EXCEL表单环境为平台，联合使用EXCEL自带的VBA编写了实用计算程序。在该程序中，子集模拟被封装在可视化的应用程序模块中。此外，边坡稳定的确定性分析( 计算边坡稳定的安全系数) 和不确定性分析分别由相对独立的计算模块实现。这种方式提高了程序的适用性，使边坡可靠度分析的过程相对清晰明了，简单易行。根据对大量算例和工程实例的总结分析，对算法进一步优化，提出Risk aggregation 的概念和算法。在项目的开展中，土工参数的统计不确定性(即由于勘察资料的有限而导致的统计误差)的影响得到充分的认识，并开发出一套贝叶斯方法，根据有限的勘察资料定量地描述土工参数的不确定性。本项目的开展及完成，推动工程可靠度分析理论的进一步发展，并在完善边坡稳定的分析理论、深化对滑坡机理的认识、滑坡治理以及滑坡预报等方面有着重要的理论价值和科学意义，同时对指导工程设计实践和保障当前各项大型工程建设的胜利完成也有着十分重要的实际工程应用价值。